The 84th Annual Meeting of the American Association of Physical Anthropologists (2015)

Beyond the Mantel test: phylogenetic mixed models and human cranial form as a multivariate response


Department of Anthropology, University of California, Davis

March 26, 2015 , Gateway Ballroom 2 Add to calendar

Attempts to quantify the extent to which morphological divergence among groups is due to neutral processes, selection, and environmental effects face a number of challenges. The partial Mantel test and phylogenetic mixed model (PMM) are two commonly used solutions. While the Mantel test has positive attributes, such as the ability to process multivariate observations, it has important weaknesses as well. Chief among these, the method only retrieves estimates of association among matrices of pairwise distances; it does not provide estimates of effect size and error in the original units of measurement. PMMs scale up quantitative genetic methods for pedigreed observations to inter- and intraspecies contexts. In contrast to Mantel tests, PMMs are designed to partition phenotypic variance among fixed and random effects. However, most PMM implementations are restricted to studies involving just a few variables, hence few parameters. Here, we apply recent innovations that extend the PMM to highly multivariate data. Our Bayesian mixed model uses an embedded factor model to estimate the phylogenetic effects matrix, the PMM analog to additive genetic effects. This simplifies the problem by reducing the number of parameters to be estimated. Using 57 linear measurements on 10 populations from the Howells craniometric data set, we replicate a previously published analysis of the relative influence of population history and climate on modern human cranial variation. Our preliminary results indicate that both population history and climate influence cranial form, though population history effects are much stronger. Most encouraging, the factor model readily accommodates the high-dimensional data set.

This study was funded with generous support from the Wenner Gren Foundation (Dissertation Fieldwork Grant) and the National Science Foundation (DDIG; Award No. BCS-1232590).